experiments in water cooling a mach

[bazx]

tom that gauge is still the broken one from the last time we talked

but the good news is i have a new one still in its box

i truly wished i understood what you just said

i am only an observer

i try things some times they work some times they dont

but i take forward the things that do work into my builds

water cooling in my units is still in its early days there is much

more here for me to observer

 

[bazx]

Would chilled water give any greater improvements or would it cause to many condensation issues?

i am being careful here that i do not create a mini cascade

this is just a mach with a passively water cooled condenser

if i use anything to chill the water it will become a cascade imo

 

[bazx]

Any floodback issues Bazx ? I think it will be a totally different ballgame though to run that on a rad as opposed to what you have now.Nothing more than a single rad could be fitted in a standard mach box and i doubt that will offer much improvement over a decent condenser.As for chilled water,if you're gonna use a 2nd compressor to chill the water,might as well do a cascade,IMO.I too have had thoughts that a lot of these overpriced evaps offer very little performance advantage over the stock mach head,hence my using a stock unit for 24/7.Good stuff guys,gotta love this shlt.

this is an intresting point

the unit is working in a way that i have not seen before

with this type of condensing you get very good temps with a much reduced charge

i will need to run more tests before i say the static charge

but the answer no flood back

 

[maverik-sg1]

^^^^^

What Tom said - over my head but made sense as well, am i right in assuming screw comps are more efficient in true displacement?

Thickies comments are right - if u change the head of the mach its only for one reason - easier mounting, the difference in temps is not enough to get you a better clock, literally at sub zero temps have to double (or is that half as we are talking minus temps) - which is why we see benching goups like this:

Air - ambient plus 28degs or higher

Water - ambient plus 15degs or more

Single Stage @ -30 core temps

Dry Ice @ -65 core temps

Cascade @-90 or lower

LN2 @ -180 or lower

All above is totally ball park but reasonable to assume.

What these sort of mods do is they help satisfy the difference between load temps and idle which offers stability over time.

What about using an external kooling system such as a Koolance for this type of mod? - The unit happily sits on top of the case - just requires pumbing and a molex then

 

[SoddemFX]

tom that gauge is still the broken one from the last time we talked

but the good news is i have a new one still in its box

i truly wished i understood what you just said

i am only an observer

i try things some times they work some times they dont

but i take forward the things that do work into my builds

water cooling in my units is still in its early days there is much

more here for me to observer

No worries i thought you'd fixed it, it confused me he he

If i can help just drop me a PM, email, phone etc but it's looks like you're doing great anyway

Tom

 

[K404]

the difference in temps is not enough to get you a better clock, literally at sub zero temps have to double (or is that half as we are talking minus temps) - which is why we see benching goups like this:

Air - +50?

Water +30- +45

Single Stage @ -30 core temps

Dry Ice @ -60 ishh

Cascade @-90 or lower

LN2 @ -130 ishh

fixed.

CPU scaling is a linear effect based on absolute temperature (Kelvin, not celcius) and voltage. (or very close to it)

Think this is off-topic though. I have a sketchy formula that is true to within estimation errors regarding the whole lot, its interesting to see it pan out

 

[sacha35]

What about using an external kooling system such as a Koolance for this type of mod? - The unit happily sits on top of the case - just requires pumbing and a molex then

This is what me and baz are working towards mav, you have a bencher that just uses a water setup he has all the rads blocks ECT, all he would need to do is take of his CPU waterblock and run the pipes from his pre-installed rad straight into some fast fix fixings that will be on the back of the Single stage.

then all he has to do is power it up and away he goes, same amount of noise as his water setup before but with the extra cooling power of a single stage, i must admit i was amassed just how fast the unit pulled down to -50 it took know more than 30 seconds.

 

[bazx]

^^^^^

What Tom said - over my head but made sense as well, am i right in assuming screw comps are more efficient in true displacement?

Thickies comments are right - if u change the head of the mach its only for one reason - easier mounting, the difference in temps is not enough to get you a better clock, literally at sub zero temps have to double (or is that half as we are talking minus temps) - which is why we see benching goups like this:

Air - ambient plus 28degs or higher

Water - ambient plus 15degs or more

Single Stage @ -30 core temps

Dry Ice @ -65 core temps

Cascade @-90 or lower

LN2 @ -180 or lower

All above is totally ball park but reasonable to assume.

What these sort of mods do is they help satisfy the difference between load temps and idle which offers stability over time.

What about using an external kooling system such as a Koolance for this type of mod? - The unit happily sits on top of the case - just requires pumbing and a molex then

i also believe that to much has been made of evaps

i think this unit would be fine with its stock evap put back on

i have one evap that i keep and use when testing units

as it is of known ability with different cap length and refrigerants

this is the one thats on there now and it has worked the way i expected it to

which is important when you need to test things

i have played around with a few ways to cool the water my favorite was a water

fountain in the garden

another i let the water drop from a height of around 2 foot onto a piece of wood floating in the tank as the water hit the wood the water splashed out

in droplets cooling its self as it flew thro the air hitting the walls of the tank and then doping into the tank

i dont see there being a problem with the water cooling tbh these little

n11 dont seem to heat the water much

and still think it could be done with out fans

with maybe the water being run down the outside of a 2 foot high metal pole that was say star shaped in profile sitting in a small res

as the water runs down the colum it would be cooled collected at the base and returned to the unit

garden centres are fully of these lil water features

 

[SoddemFX]

^^^^^

What Tom said - over my head but made sense as well, am i right in assuming screw comps are more efficient in true displacement?

I don't want to get off topic but yes Mav you are right, in typical conditions a reciprocating compressor will be in the 40 - 50% region, a rotary or scroll will be in the 70% region. It's just the design on them, i'm not sure what makes the rotaries/scrolls higher but they are. The highest volumetric efficiency (in these conditions) reciprocating compressor i've seen is the Danfoss NL7CLX which IIRC is about 75% but that's abit of an odd ball, most fall in the 40-50% region and generally for reciprocating compressor the bigger the displacement the lower the volumetric efficiency.

On the big recips (40cc+) you'll struggle to get 40% at 10:1, again i don't know why the large compressors tend to have lower volumetric efficiency but they do.

Rotaries just hold that 70% figure right over the working compression ratio range, but i don't know how about that either

Tom

 

[bazx]

few more tests last night before i shorten the cap tube

32m @ 4600mhz with 1.7v on the same chip 6800

Time unit was running 1 .5 hours

Water barrel temp before 16.1c

Water barrel at the end 19.5c

Barrel size 120l

Ambient in shed 15c

Ambient outside shed 12c

the tests were run around 8pm in the evening so the ambient temps had dropped quicker the the water temps

as you can see the unit is running under load in a vacuum

this however is where the best loaded temps were to be had -47

when i increased the charge so the unit was running @ 0 vacuum the temp went up by 4c so -43 under the same load

the high pressure stayed @ 13 bar

 

[bazx]

loaded hx temps 32m@4600mhz/1.7v

HX temps peaked at T1 36c measured at the start of the hx and T2 21c as the water left the hx

at the start of the run as image

 

[bazx]

 

[sacha35]

very impressive mate, your getting some great results with the setup.

 

[Thickbrit]

I was able to reach the same clocks (4596) for 32m on that x6800 on my home brewed custom 507 s/s with only 1.66v,FWIW.

 

[bazx]

so far the water cooling has work in that it reduced the high side

pressures in the unit at this load by quit a bit

which means to me that i can put in a colder refrigerant now

i think this is where the real gains will come from now

 

[SoddemFX]

Hi Baz,

Great work again

I know it will only be an indication but what is the temperature of the copper pipe at the exit of the suction line, towards the end of the 32M calculation?

Is 36C the discharge temp of the compressor?

Under load towards the end of the 32M calc, was the discharge pressure still 13Bar gauge?

Out of interest with a water butt volume of 120l (120Kg or near enough), a specific heat capacity of water being 4181 J Kg^-1 K^-1 and a temperature rise of 3.4K (19.5C - 16.1C) you have put 120 x 4181 x 3.4 = 1705848J of energy into the water butt.

As this was over a 1.5hour period (5400 seconds) this means your watercooled condensor is dissipating 1705848 / 5400 = 316W from the system. This is your condensor load and it's roughly in the range of what i'd expect.

This calculated load figure might be a little higher than in reality because the water in the tank won't have mixed entirely and some parts of the water butt will be colder than others, the butt will "leak" heat also so the system isn't totally enclosed but i'd be supprised if it was +/- 10% either way tho

Tom

//Edit - specific heat capacity will be more like 4184 at the temp of the butt but it doesn't make much difference at all...

 

[bazx]

Hi Baz,

Great work again

I know it will only be an indication but what is the temperature of the copper pipe at the exit of the suction line, towards the end of the 32M calculation?

Is 36C the discharge temp of the compressor?

Under load towards the end of the 32M calc, was the discharge pressure still 13Bar gauge?

Out of interest with a water butt volume of 120l (120Kg), a specific heat capacity of water being 4182 J Kg^-1 K^-1 and a temperature rise of 3.4K (19.5C - 16.1C) you have put 120 x 4181 x 3.4 = 1705848J of energy into the water butt.

As this was over a 1.5hour period (5400 seconds) this means your water cooled condenser is dissipating 1705848 / 5400 = 316W from the system. This is your condenser load and it's roughly in the range of what i'd expect.

This calculated load figure might be a little higher than in reality because the water in the tank won't have mixed entirely and some parts of the water butt will be colder than others, i'd be surprised if it was +/- 10% either way tho

Tom

i tried to get as much info down as pos with this set

the 36c is just as it enters the the hx which is around 10cm from the compressor

so yes 36C is discharge temp of the compressor

the problem is now tonight the water temp was a fair bit lower due to weather conditions and the temps on the discharge did not get above 24c

with the temps exiting the hx at 16c

out of interest i added a small amount of r23 to bring the high side up to 15 bar

this had the effect of decreasing the temps on the same load further to -49

which the hx held rock solid

so i feel sure i could get away with another gas

tom witch gas would condense at the current hx temps

 

[bazx]

discharge pressure still 13Bar gauge?

just read it again missed this

the 13 bar is the high side pressure mesured after the hx /condenser

this pressure does not change at all at load or idle

 

[SoddemFX]

That's interesting, thanks Baz

Personally i wouldn't blend r23 with another gas for a single stage but i think there is a commercial blend of r404a and r23 called RS23 or something like that so it can work... maybe look it up if you want to try it...

You're currently condensing at 28C so r410a would condense at about 17Bar gauge, so expect to see in the region of 245PSI gauge. With the higher compressor load from higher pressure r410a, the load on the condensor will increase so maybe a bit higher.

If you're running around the same suction pressure (1Bara, 0Barg) then that compression ratio will be getting quite heavy, wether you want to try it is obviously your choice - if you have r410a i think you should try for the hell of it Just keep and eye on the discharge temperatures and pressures when you're charging.

To be honest you're pushing the limits of what a single stage can do, there isn't that much more you can achieve without cascading it

Tom

//Edit - fixed error in pressures

 

[bazx]

the suction pressure with the 23 added edged up to 0 but not too it

still in slight vacuum with h/s @ 15 bar

i also would not want to go with any kind of blending myself as i would have no way of getting it the same again

from what i can tell the 23 did not mix any ho

just acted to increase the pressure in the unit

just like when i used co2 with 402 in cascade 1